The result of this increased fitness cost was substantially better therapeutic efficacy of AAV-vectored VRC07 antibody, because the barrier to escape had been raised.

In experiments testing this pol mutation in combination with env escape mutations in envelope, we found that there was no impact on resistance benefit, but substantial increases in fitness cost, particularly with either combination of the two envelope mutations needed to escape.

To understand the contribution of viral fitness to VRC07 escape, we performed a similar experiment in mice that had been previously treated with escalating doses of antiretroviral drugs to induce escape, which occurs through a point mutation in HIV polymerase that lowers fitness.

To explore the contribution of sequence context to VRC07 escape, we created chimeric strains between the original strain and the second strain and found that sequences within the V5 loop seemed to dictate whether mutations in the D loop would result in partial or complete escape.

We then repeated these experiments with another strain and found that in contrast to the first experiment, none of the antibodies could suppress HIV replication. This was because individual mutations in this strain background enabled complete escape at minimal fitness cost.

By integrating the area under the curve for each path and scaling this by the frequency with which we observed each mutation pattern, we were able to quantify an escape barrier score for each antibody and found that VRC07 posed a dramatically higher barrier than other bNAbs used.

We used these measurements of fitness cost and resistance benefits to generate "escapability" maps, showing the path taken by viruses during escape from bNAbs. This allowed us to visualize the much harder escape path posed by VRC07.

The escape mutations for N6 and PGDM had fairly limited fitness costs, but VRC07 escape mutations imposed significant penalties to viral growth.

After engineering these mutations back into the parent strain and testing their relative resistance to neutralization we found that individual mutations led to complete escape from N6 of PGDM1400, but VRC07 required at least two mutations.

When we sequenced the mutations that arose in the envelope gene from all of the mice exhibiting viral escape, we found that the virus had acquired at least two mutations in the case of VRC07, but only one mutation in the case of N6 or PGDM1400.